Radiative Transfer Theory at Optical wavelengths applied to vegetation canopies: part 2 UoL MSc Remote Sensing Dr Lewis [email protected]
Mar 28, 2015
Radiative Transfer Theory at Optical wavelengths applied to vegetation canopies: part 2
UoL MSc Remote Sensing
Dr Lewis [email protected]
Radiative Transfer equation
• Used extensively for (optical) vegetation since 1960s (Ross, 1981)
• Used for microwave vegetation since 1980s
Radiative Transfer equation
• Consider energy balance across elemental volume
• Generally use scalar form (SRT) in optical• Generally use vector form (VRT) for microwave
z
Pathlength l
z = l cos l
Medium 1: air
Medium 2: canopy in air
Medium 3:soil
Path of radiation
Scalar Radiative Transfer Equation
• 1-D scalar radiative transfer (SRT) equation– for a plane parallel medium (air) embedded with a low
density of small scatterers– change in specific Intensity (Radiance) I(z,) at depth z
in direction wrt z:
( ) ( )zJzIz
zIse ,),(
,Ω+Ω−=
∂
Ω∂κμ
Scalar RT Equation
• Source Function:
• - cosine of the direction vector (with the local normal
– accounts for path length through the canopy
• e - volume extinction coefficient
• P() is the volume scattering phase function
( ) ′′→′= ∫ dIzPzJ S ),();,(,4π
( ) ( )zJzIz
zIse ,),(
,Ω+Ω−=
∂
Ω∂κμ
Extinction Coefficient and Beers Law
• Volume extinction coefficient:– ‘total interaction cross section’– ‘extinction loss’– ‘number of interactions’ per unit length
• a measure of attenuation of radiation in a canopy (or other medium).
e
( ) leeIlI κ−= 0Beer’s Law
Extinction Coefficient and Beers Law
€
I l( ) = I0e−κ e l
I
eIdl
dI
e
le
e
−=
−= −
No source version of SRT eqn( ) ( )zJzI
z
zIse ,),(
,Ω+Ω−=
∂
Ω∂κμ
Optical Extinction Coefficient for Oriented Leaves
• Volume extinction coefficient:
• ul : leaf area density – Area of leaves per unit volume
• Gl : (Ross) projection function€
e Ω( ) = ulGl Ω( )
( ) ( ) lllll dgG ΩΩ⋅ΩΩ=Ω ∫+ππ 22
1
Optical Extinction Coefficient for Oriented Leaves
( ) ( ) lllll dgG ΩΩ⋅ΩΩ=Ω ∫+ππ 22
1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
G_l(theta)
zenith angle / degrees
spherical planophile erectophileplagiophile extremophile
Optical Extinction Coefficient for Oriented Leaves
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
G_l(theta)
zenith angle / degrees
spherical planophile erectophileplagiophile extremophile
• range of G-functions small (0.3-0.8) and smoother than leaf inclination distributions;
• planophile canopies, G-function is high (>0.5) for low zenith and low (<0.5) for high zenith;
• converse true for erectophile canopies;• G-function always close to 0.5 between 50o and 60o
• essentially invariant at 0.5 over different leaf angle distributions at 57.5o.
Optical Extinction Coefficient for Oriented Leaves
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
G_l(theta)
zenith angle / degrees
spherical planophile erectophileplagiophile extremophile
• so, radiation at bottom of canopy for spherical:
• for horizontal:
€
−ulG Ω( )
μdz =
LG Ω( )μz= 0
z=− H
∫
( )
LLG
eIeI5.0
00
−Ω
−
=
LeI −=
A Scalar Radiative Transfer Solution
• Attempt similar first Order Scattering solution– in optical, consider total number of interactions
• with leaves + soil
• Already have extinction coefficient:
( ) ( )= Guleκ
SRT
• Phase function:
• ul - leaf area density;
• ’ - cosine of the incident zenith angle• - area scattering phase function.
( ) ( )→′′
=→′ luPμ
1
SRT
• Area scattering phase function:
• double projection, modulated by spectral terms l : leaf single scattering albedo
– Probability of radiation being scattered rather than absorbed at leaf level– Function of wavelength
€
′→ ( ) =1
4πωlgl Ω l( )Ω ⋅Ω l ′ Ω ⋅Ω l dΩ l
4π
∫
SRT
( )
( ) ( )
( ) ( )
( )( ) ( )
( ) ( )
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−
+→′
+−=
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
+
0
00
0
00
1
,
,
00
0
000
μμ
μμ
μμ
μμ
μμ
δρ
s
ss
s
ss
GGL
ss
sssoil
GGL
s
eGG
I
Ie
zI
SRT: 1st O mechanisms
• through canopy, reflected from soil & back through canopy
( )
( ) ( )
( ) ( )
( )( ) ( )
( ) ( )
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−
+→′
+−=
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
+
0
00
0
00
1
,
,
00
0
000
μμ
μμ
μμ
μμ
μμ
δρ
s
ss
s
ss
GGL
ss
sssoil
GGL
s
eGG
I
Ie
zI( ) ( )
( )0,0
00
ΩΩ⎟⎟⎠
⎞⎜⎜⎝
⎛ Ω+Ω−
ssoil
GGL
s
ss
e ρμμ
μμ
1. 2.
SRT: 1st O mechanisms
( )
( ) ( )
( ) ( )
( )( ) ( )
( ) ( )
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−
+→′
+−=
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
⎟⎟⎠
⎞⎜⎜⎝
⎛ +−
+
0
00
0
00
1
,
,
00
0
000
μμ
μμ
μμ
μμ
μμ
δρ
s
ss
s
ss
GGL
ss
sssoil
GGL
s
eGG
I
Ie
zI
1. 2.
( )( ) ( )
( ) ( )
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−
+→′ ⎟⎟
⎠
⎞⎜⎜⎝
⎛ +−
0
00
100
μμ
μμ
μμs
ss GGL
ss
eGG
Canopy only scattering
Direct function of Function of gl, L, and viewing and illumination angles
1st O SRT
• Special case of spherical leaf angle:
( )
( ) ( ) γργγγπτρ
cos3
cossin3
5.0
lll
G
+−+
=Ω→Ω′Γ
=Ω
′⋅=γcos
Multiple Scattering
LAI 1
Scattering order
Contributions to reflectance and transmittance
Multiple Scattering
LAI 5
Scattering order
Contributions to reflectance and transmittance
Multiple Scattering
LAI 8
Scattering order
Contributions to reflectance and transmittance
Multiple Scattering
– range of approximate solutions available– Recent advances using concept of recollision
probability, p• Huang et al. 2007
Q0
s
i0
i0=1-Q0
p
s1=i0 (1 – p)
p: recollision probability: single scattering albedo of leaf
• 2nd Order scattering:
pi
s
i
s
1
2 =
i0
i0 p
2 i0 p(1-p)
( ) ( ) ( ) L+−+−+−= 232
0
111 pppppi
sωωω
( )pi
s−= 1
0
1 ω
( ) ( ) ( ) L+−+−+−= 232
0
111 pppppi
sωωω
( )[ ]L+++−= 22
0
11 pppi
sωωω
( )
p
p
i
s
−
−=
1
1
0
‘single scattering albedo’ of canopy
( )
p
p
i
s
−
−=
1
1
0
( ) ( )λλ
pns −
=1
1 Average number of photon interactions:The degree of multiple scattering
p: recollision probability
( ) ( )( )λλλα
p−
−=
1
1Absorptance
Knyazikhin et al. (1998): p is eigenvalue of RT equationDepends on structure only
• For canopy:
0.00.10.20.30.40.50.60.70.80.91.0
0 2 4 6 8 10
LAI
pcanopy
Smolander & Stenberg RSE 2005
( )( )bcanopy kLAIpp −−= exp1max
pmax=0.88, k=0.7, b=0.75Spherical leaf angle distribution
Canopy with ‘shoots’ as fundamental scattering objects:
canopyi
s⎟⎟⎠
⎞⎜⎜⎝
⎛
( )shootcanopy
shootcanopycanopy p
p
ω
ωω
−
−==
1
1
Clumping
( )shootcanopy
shootcanopycanopy
canopyp
p
i
s
ω
ωω
−
−==⎟⎟
⎠
⎞⎜⎜⎝
⎛
1
1
0
Canopy with ‘shoots’ as fundamental scattering objects:
( )needleshoot
needleshootshoot
shootp
p
i
s
ω
ωω
−
−==⎟⎟
⎠
⎞⎜⎜⎝
⎛
1
1
0
( )needle
needlecanopy
canopyp
p
i
s
ω
ωω
2
2
0 1
1
−
−==⎟⎟
⎠
⎞⎜⎜⎝
⎛
( ) shootcanopycanopy pppp −+= 12
p2
pcanopy
Smolander & Stenberg RSE 2005
• pshoot=0.47 (scots pine)
• p2<pcanopy
• Shoot-scale clumping reduces apparent LAI
Other RT Modifications
• Hot Spot– joint gap probabilty: Q
– For far-field objects, treat incident & exitant gap probabilities independently
– product of two Beer’s Law terms
′′′
→′
)G( + )G( L -
= )Q( e
RT Modifications
• Consider retro-reflection direction:– assuming independent:
– But should be:
)G( L 2
-
= )Q(Ω
Ω→Ω e
)G( L
-
= )Q(Ω
Ω→Ω e
RT Modifications
• Consider retro-reflection direction:– But should be:
– as ‘have already travelled path’– so need to apply corrections for Q in RT
• e.g.
)G( L
-
= )Q(Ω
Ω→Ω e
),C( )P( )P( = )Q( ′′→′
RT Modifications
• As result of finite object size, hot spot has angular width– depends on ‘roughness’
• leaf size / canopy height (Kuusk)• similar for soils
• Also consider shadowing/shadow hiding
Summary
• SRT formulation– extinction– scattering (source function)
• Beer’s Law– exponential attenuation – rate - extinction coefficient
• LAI x G-function for optical
Summary
• SRT 1st O solution– use area scattering phase function– simple solution for spherical leaf angle– 2 scattering mechanisms
• Multiple scattering– Recollison probability
• Modification to SRT:– hot spot at optical